Project Report No. 88

REPORT ON

PROPERTIES OF COHESIVE SOILS RELATED TO EROSION

by

John W. Hayden

Prepared for

NATIONAL SCIENCE FOUNDATION

under

Grant GK-364

January 1967

REPORT ON

PROPERTIES OF COHESIVE SOILS RELATED TO EROSION

by John \~. Hayden

ABSTRACT

During the past year, an investigation has been underway on the

physical $t.nci erosive properties of cohesive soils. The physical properties,

including tensile strength, of four cohesive soils have been determined.

It was found that the tensile strength of the cohesive soils tested lies

between the compression strength at 1 per cent strain and the compression

strength at 2 per cent strain as determined by an unconfined compression test.

Thus, the tensile strength of a cohesive soil is no better as a predictor of

erosion characteristics than the unconfined compression strength, and neithe~

is a good predictor of critical shear stress required to start erosion~

The variation of the erosion oharacteristics with moisture content for

the clay of highest plastic index (PI; 33) was determined by means of flume

tests. Both the critical shear stress and the erosion rate were measured.

The experimental data indicate that as the moisture content of the soil was

increased from approximately 40 per cent to 60 per cent, the critical shear

stress increased and the erosion rate decreased~ With a further increase in

moisture content, critical shear stress decreased with an accompanying in~

crease in the erosion rate. For moisture contents below 40 per cent and abov~

the liquid limit (IJ?6S), the critical shear stress approached zero, and at

moderate shear stresses (r--0.2 Ibs/ft2), the data indicate that the erosion

rates could exceed 1000 gm/ft2/hr.

INTRODUCTION

The scour and erosion of soil due to the motion of fluid is one of the

most important problems encountered in the design of hydraulio structures.

However, because of the complexity of the problem, it is perhaps one of the

least understood. Usually, the boundary and flow conditions are constantly

changing; thus, vigorous mathematical treatment of the problem is extremely

difficult if not impossible. With model analysis usually being used to

empirically study the erosion process, additional problems are introduced.

The primary difficulties are that of modeling the sediment properly and

dOrr~QtlY interpreting the results. However, an engineer who knows the oha.~~oteristics of a noncohesive soil (mainly grain size. density, and shape characteristics) oan obtain an accurate picture of the soour pattern and transport rate for a particular problem by studying the available literature and performing a minimum amount of model testing.

For cohesive soil, only a limited. amount of basic information is available oonoerning the erosion process and transport rate. Most of the early investigations involving cohesive soil were conduoted either in relation to a speoifio hydraulio structure or for a ohannel, to determine .

the maXim~m allowable ohannel velooity below which no erosion of the bed would ocour. Although a model study serves a useful purpose, only a limited amount of basic knowledge leading to an overall understanding of the erosion process is obtained. Some notable exceptions to the above approaohes are those studies [1-11J* in whioh an attempt was made to relate the basio soil properties to the erosion oharaoteristios of the soil.

Despite the importanoe of erosion of oohesive soils, only a limited effort has been direoted toward a systematio study of the problem. Reoog~

nizing the need for a oonoentrated study of the erosion of cohesive soil by

both hydraulic and soil engineers, a research program in this direction was undertaken_

TEST PROCEDURE

The erosion tests were oonduoted in a glass-walled hydrauliC flume 6 in. wide and 15 ft long. A false floor was added to provide room for the dynamometer which measured the shear stress on the sample, The sample and dynamometer were plaoed as shown in Fig. 3 with a sluioe gate about 5 ft upstream from the sample, as it was neoessary to use superoritical velooities to provide enough shea~ stress to erode the soil. The shear stress on the sample was varied by ohanging the disoharge, with the sluioe gate being adjusted to maintain a oonstant depth of 0.05 it for all runs. Water for the tests was taken direotly from the MiSSissippi River and was relatively olear and sediment-free.

* Numbers in bra.okets refer to the List of Referenoes on page 9.

The erosion sample was prepared by kneading soil of the desired , I

moisture bontent into the sample pan in suoh a way as to remove all air pookets and large voids. The sample was then submerged and subjeoted to ~ coh'solidation of 0 .. 2 tsr for 24 hours and. allowed to rebound for another 24 hours. After rebounding, the sample surfaoe was trilll1lled and smoothed td eliminate any roughness (Fig. 4a).

The prepared sample was weighed, plaoed on the dynamometer, and adjusted so that the sample surfaoe was flush with the false floor of the channel. With the sample in plaoe, the Sanborn reoorder was balanoed and the disoharge valve opened to the desired flow. When the flow had reaohed equilibrium, the shear stress would beoome oonstant and oould be read from

the reoorder with the aid of the oalibration chart.

After a two~hour period at oonstant shear stress, the sample was

removed and weighed to determine any erosion loss. The sample was replaced and the process repeated, but at greater shear stress. After the sample had been subjected to the maximum velocity of the flume or had eroded excessively, it was removed, the final weight determined, a representative moisture sample

taken, and the sample soil stored for another run (Fig. 4b-o).

The compressive strength of the soil was determined using the unoon­

fined compression test (ASTM Des. 2166-6JT) [12J. Using a controlled rate of strain devioe, a stress-strain cur~e was obtained for each sample so that one could obtain the strength of the soil for any inorement of strain.

Sinoe very little information is available about tensile tests of cohesive soils, it was neoessary to design and. build the tensile test

equipment "from scratch." Hence, a dumbbell-shaped mold was machined from lucite plastic as shown in Fig. 5. The tensile sample was formed by forcing soil into the mold ensuring that all the la:rge voids and. air pockets were removed. After trimming the excess 80il away, the side pieces were removed and. the sample placed on the styrofoam blocks as shown in Fig. 6. One end

of the sample was attached to the spring balanoe and the other end to the loading screW. The loading soreW was then tu:rned to p:roduce a strain rate of 0.05 in. pel' minute. The sample was thus loaded to failure~ the tensile strength being the maximum load before failure. After the test, a moisture sample was taken from the failed area.

-4-

DISCUSSION

A. Tensiie and Unconfined Compression Tests

The tensile strength and unconfined compression strength were ob­

tained f.rom four Minnesota sOils with plastic jndices of 7, 12, 17, and 33. The samples had a considerable variation in characteristics from a loam to

a heavy clay. The unconfined compression tests were conducted at moisture contents from below the plastic limit to the liquid limit. Because of the nature of the tensile test, it was possible to obtain results at moisture

contents above the liquid limit of the soil. The typical mode of tensile

failure for samples in the lower plastic range was a sudden fracturing at the critical section as in Fig. 7, Near the liquid limit, the sample would

undergo a typical plastic failure with a constant or slightly decreasing

stress at large strains.

The failure strength for the unconfined compression test was taken

as the stress at 15 per cent strain or the maximum stress attained in the

sample. This is a slight deviation from the ASTM specification which defines the failure strength at 20 per cent strain; however, for most samples, the st:r-ess had become constant at a strain of 1.5 per cent.

The plots of unconfined compression strength and tensile strength as

a function of moisture content for the four soils are shown in Fig. 8 through

Fig. 11. Also included are plots of the unconfined compression stJ:length

for stl'ains of 1 per cent and 2 per cent. The tensile strength of the

three more plastic soils is seen to fall between the compressive strength

at 1 per cent and 2 peJ:l cent strain, while the tensile strength of the low

plasticity soil (PI = 7), shown in Fig. 8, falls very much below the 1

per cent unconfined compression stJ:length.

The unconfined compression strength of a soil can be expressed as [13J:

<lu = 2c JNj

where c is the cohesion of the soil, and

( 2)

\.

.. .5-

and ¢ is the angle of internal friction of the soil.

If it is assumed that for most clays, ¢ will equal zero, then

0:= ~ 2

HencH~ ~ the cohesion of a olay soil will be equal to half its unoonfined

compression strength.

The tensile strength. of a soil can be approximately related to

cohesion by the ~quation [14J:

':r := -2c

and since

c = f T:= -~

(4)

(3)

(.5 )

or the tensile strength of a soil will be approximately equal to its unoon. fined compression strength. ':rhus, for soils of moderate plasticity the equation

T:= -~ (.5 )

will be approximately correct if the value for the unconfined compression stl"ength is taken at a strain between 1 per cent and 2 per cent. On the

basis of these findings, one would oonolude that the tensile strength of a

soil would have little advantage over the unoonfined oompression strength

for predicting erosion.

It may be argued that the results of the unoonfined compression or

tensile tests are not directly applicable to the ero.sion tests, because the

erosion samples were subjected to a consolidation not used in the other tests. However, the unconfined compression and tensile tests were made to determine

their relationship to one another and to determine their general relationship to the erosion of the soil. A consolidation will cause the strength vs.

-6-

moisture content curves to be shifted slightly but the general shape of

these curves will remain the same. Thus, no attempt is made to apply these

tests quantitatively to erosion.

B. Erosion

The erosion tests were conducted on only one of the soils, a Duluth red clay, which has a liquid limit of 65 and a plastic index of 33~ The

samples Were eroded at moisture contents from 44 per cent to 74 per cent,

which would correspond to the upper plastic and liquid states of the soil.

At values below 44 per cent, it was difficult to produce a uniform sample with the small amount of consolidation which was used.

The more resistant samples would have no erosion at the lowest shear

stress values and would erode in small particles at random points on the

sample surface at the highest shear stress values as shown in Fig~ 4b. The

samples with lower resistance would undergo a mass failure of the surface

as shown in Fig. 4c. It was impossible to leave these latter samples in the

flume for the full two-hour period, as they would be completely destroyed.

Hence, after about 10 per cent of the sample had been eroded, the run was

stopped. The shear stress for these runs of high erosion was measured at the

beginning of the run before appreciable erosion had occurred.

If one compares the work done by Lyle and Smerdon [8J with that done

by Grissinger [7J (see Figs. 1 and 2), the investigators appear to be in

disagreement about the effect of void ratio or moisture content (for a

saturated soil, moisture content equals void ratio divided by the specific

gravity of the soil)~ Lyle and Smerdon found that erosion resistance de.

creases with an increase in void ratio (moisture content), while Grissinger

stated that erosion resistance will increase with an increase in moisture content (void ratio). However, it is interesting to note one of Grissinger's

curves shown in Fig. 1. The curve for samples packed at 17 per cent moisture

with no vacuum treatment, like the other curves, shows an erodibility de­

oreaseas the moisture content approaches the plastic limit, but as the

moisture content approaches the liquid limit, the erodibility is seen to

increase. The erosion tests of this investigation have shown similar

behavior whioh would appear to make the conclusions of both studies correct.

-7-

E~osion resistance was found to increase with moisture content up toa moisture content of approximately 55 per cent and then to decrease

as moisture content continued to increase. This can be seen by examining

Fig. 12 which is a plot of shear stress vs. moisture content. On this plot,

an envelope line can be drawn which encompasses all points of zero erosion.

This line is the locus of all points which are just on the verge of erosion

or points which are at the critical tractive stress. From the plot, the

maximum oritica1 tractive stress is seen to be 0,15 psf or about 0.001 psi.

The 60i1 9s shear strength at this moisture content is approximately half its

unconfined compression strength or about 0.25 psi. (Had the unconf.ined com­

pression tests been oonducted with consolidation, the shear strength would

be somewhat higher [14].) Thus, the soil failed at about two orders of

magnitude below its shear strength; this is similar to the findings reported

by Berghager and Ladd [lJ. Henoe, erosion behavior cannot be predicted

from a strength analysis of the soil, because erosion failure does not

oocur at the same order of magnitude as the soil strength and beoause the

critioa1 tractive stress does not follow the same pattern as the unconfined

compression strength of the soil. At the liquid limit, tensile and com­

pression strength are small so that the erosion resistanoe should be small,

which the tests have shown to be true. At lower moisture contents, the

strength of the soil is high so that it should have a high erosion resistance.

Howev'er, the tests have shown that the erosion resistance at low moisture contents is also low.

It may be suggested that this behavior is the result of density

variations in the sample; however, a plot of wet density vs. moisture

content (Fig. 1)) for the samples shows this is not the case. The density

is seen to gradually decrease with an increase in moisture content, as would

be expected.

If one considers the limiting oonditions for the soil, the erosion

behavior can be partially explained. When the soil is at zero moisture

content, it will behave as a oollection of discrete particles with very low

erosion resistance. When the soil is at high moisture contents, well above

the liqUid limit, it will behave as a liquid and have low erosion resistanoe.

But between these extremes, the soil will behave as a two-phase system with

-8-

the $oil and water interacting to produce a cohesive mass of varying

erpsion resistance. It is reasonable to assume that somewhere in this area

of two-phase behavior, the erosion resistance will be a maximum and that

near the limiting conditions, the resistance will be smaller as the tests

have shown. However, it is difficult to explain the position of the maximum

point in the upper plastio range. The position of this point is probably

dependent upon the amount of pre consolidation present in addition to several

other variables, so that it is best predioted from experimental observations~

CONCLUSIONS

On the basis of these stUdies, several conclusions may be drawn.

Since the tensile strength of a 60il follows essentially the same pattern

as the unoonfined compression strength, it offers no advantage over the

unconfined compression strength for predicting soil erosion. The tensile

tests have shown that for a soil of moderate plasticity, the tensile

strength will be approximately equal to the unconfined compression strength

at a strain between 1 per cent and 2 per cent.

The erosion stUdies have shown that the critical tractive stress

variation with moisture content for a given soil bears little relation to

its variation in tensile or unconfined compression strength with moistUre

content. The critical tractive stress was found to increase with moisture oontent up to a point and then to decrease as moisture content 'Was further

increased.

-9-

LIST OF REFERENCES

[lJ Berghager, D. and Ladd, C., "Erosion of Cohesive Soils," Research Report R64-1, School of Engineering, Department of Civil Engineering, Massachusetts Institute of Technology, January 1964.

[23 Carlson, E. J. and Enger, P. F., Tractive Force Studies of Cohesive Soils for Design of Earth Canals, Bureau of Reclamation No. HYD .504, 1962.

[33 Dunn, I. S., "Tl'active Resistance of Cohesive Channels," Journal of ~, Vol. 85~ No. SM3, June 19.59.

[4J Einstein, H. A. and Krone, R. B., Flume Studies of the Transport of Sediment in Estuarial Shoaling Processes, Final Report, Hydro. Engineering Lab. and San. Engineering Res. Lab., University of California, 1962.

[5J Engel', P. F., Canal Erosion and Tractive Force Study-Analysis of Data Taken on a Boundary Shear Flume, Bureau of Reclamation No. HYD .532, 1963.

[6J Espey, W. H., A ew Test to Measure Scour of Cohesive Sediment,· Technical Report HYD-Ol- 309, HYd. Eng. Lab., Department of Civil Engineering, University of Texas, Austin, Texas, April 1963.

[7J (lris singer , E. H., "Resistance of Selected Clay Systems to Erosion by Water," Water Resources Research, American Geophysical Union, Vol. 2, No.1, 1966.

[8J Lyle, vl~ M. and Srnerdon, E. T., "Relation of Compaction and Other Soil Properties to the Erosion Resistance of Soils," Transactions ~, Vol. 8, No.3, 196,5.

[9J 1-190re , W .. L. and Hasch, F. D., "Experiments. on the Scour Resista.nce of Cohesive. Sediments," Journal of Geophysical. Research, Vol. .. 67, Ho ~ 4, April 1962.

[10J Parthenidides, E., "Erosion and Deposition of Cohesive Soils," f!:2.­ceedings of ASCE, Vol. 91, No. HIl, January 1965.

[llJ Smerdon, E. T. and Beasley, R. P., "The Tractive Force Theory Applied to Stability of Open Channels in Cohesive Soils," Research ]3ulletin 712, College of Agriculture, Agriculture Experiment Station, Columbia, Missouri, October 1959.

[12J American Society of Testing Materials--Standards 1964, Part 11, "Bituminous Materials, Soi1s~"

[lJ]

[14J

.. 10 ..

Terzaghi, K. and Peck, Ret Soil Mechanics in Engineering Practice, Wiley_ New York, 1948.

Hough, B. K., Basic Soi~s Engineering, Ronald Press, New York, 1957.

~ QE FIGURES

Figure

1 Influences of Packing Technique and Ant~cedent Water on the Erodibility of a Grenada Silt Loam at a 1.4 Bulk Density (after Grissinger)

2 Critical Tractive Force Vs. Void Ratio (after Lyle and Smerdon)

3 Schematic Diagram of Erosion Flume

4 Prepared Sample

5 Tensile Mold

6 Tensile Apparatus

7 Failed Sample

8 Tensile and Unconfined Compression Strength Vs. Moisture Content (Plastic Index = 7)

9 Tensile and Unconfined Compression Strength Vs. Moisture Content (Plastic Index = 12)

10 Tensile and Unconfined Compression Strength Vs. Moisture Content (Plastic Index = 17)

11 Tensile and Unconfined Compression Strength Vs. Moisture Content (Plastic Index = 33)

12 Shear Stress Vs. Moisture Content for Erosion Tests of Duluth Red Clay (Plastic I~dex == 33)

13 Density Vs. Moisture Cont-ent for lTIrosion Samples (Plastic Index == 33)

C

E

E c;; ,-::1 '" 0

lLJ

0

I c

6(

SCI ~ \ I \ \

'" '+l.

3:.-

20

" I ;~, ~"

.5 1 C'l 15 20

" \ \ \

• , , ., , , , \ \ \

, \ , , , ,

-,

25

\~ \

e'\Jt , , I , , I

30

Antecedent p, oisturc f Per Cent

• '~

• Packed at 10';-> moisture

Packed at 12;<:' moisture

Packed at 1 T>~ moisture

Vacuum treated to obtain

maximum particle orientation

Plastic i 1m it =- 20

Liquid i imit = 31

35 40

Fig. - InfluencE's of Packing Technique and Antecedent WCter on the Erodibility of a Crenada Silt Loam at a 1.4 Bulk Density (After Grissinger)

45 .

Approximate Moisf'ure Content, Per Cenl'

0.0332~2 ____ ~30 ______ 3~7 ______ 4~5 ______ 5~2 __ ~~~~"9 ______ ~67 ______ ~75

0.031

0 0 029

0 0 027

N d:: 0 0 025 ~

Q) U 1-0 0 u..

~ 0.023 .... u 0 1-0 t-

o 0.021 u .... ·c V

0 0 019

0.8 1.0 1.4

Void Ratio

Fig" 2 - Critical Tractive Force Vs Void Ruth (After Lyle and Smerdon) . ;

Riv¢r .. we':!::i supply

Heac: tank

FalSE: f\ oor

Supply valve

1 I '+

51 uice sate

5'

SCr.1P I c and dyncr::olT'ctcr

Fig. 3 - SchcrrJctfc D;ag,om of Erosion Flume

I Vaste

Tai:watc: control gGte

Erosion ra1'e := 1910 gm/ftt'jhr Shear stress := .25 psf Flow is left to right

6 Tens! ~e apparotus

Fig. 7 Failed sample

N c ~ ...0

, ...r. .....

Q) C (lJ L.. ....

30'1 ----------.----------.--=------,r---------,----------.--------~._--------_,

24,1 -------r------~--+---~------~

2~';

I 18

I \ I r-- 15°'~

1 0/ 1 ;0

T 15% compressive strain

• • • 2°{; compressive strain

1°'; compressive strain

Tensile strength

P I as tic lim it = 1 9

Liquid li!'Y1it::- 26

Vl 12

61 9\1' 1\ , i: '\ L: '\:

T8nsile

01 " • P"?l o 5 10 15 20 25 30 35

Iv'oisture Content r Per Cent

Fig. 8 - Tensile and Unconfined Compression Strength Vs iv':oisture Content (Plastic Index = 7)

N .~

::n-~

. .!: .... 0) C G.: '-....

V)

15,r----------,-----------.-----------r-----------,----------,-----------.-----------,

12~1 --------~-------+--------~------~

9

6

3 1 .... ... '-

T 15% compressive strain

• A

• 2<:;; compressive strain

1 ~'; compr2ssivc strain

T2:1sil::: strength

P I as tic I i;on ; t == 1 8

liqui:l limit:= 30

~

'11 I 1 swb • I 1

!5 20 25 30 35 40 5 In v

J\/,oisture Content, Per Cent

Fig. 9 - Tznsile ana ,_!n;:on:incd Compression Str.:3ngth Vs ""oisture Content (Plastic Index =: 12)

N s::

'''-. ..n

... ...r::. .....

C) s:: 0 ....

of-VI

15~i ----~----~----~----~----~----~--~

12~1 --------~--------~--------+-------~

9

6

31 1'-:",' "" 1 \:

T i5S;) COr.lFessive stroin

• A

• 20;~ compressive strain

1 o,~ compressivE: strain

Tens i I E:: strength

Plasti~ limit:::. 29

Li'"1uid I im it =- 46

0 1 -.,Z I' 45 15 20 25 30 35 40

Moisture Content, Per Cent

50

Fig, 10 - Tensile and Unconfined Compression Strength Vs Moisture Content (Pi?stic Index =: 17)

N c

':::. ..Q

.. ..£: .0-

m c I!> '-....

V)

lS~--~----It----~----r---~----~----~--~----l

.... 15C:i~ compressive strain • ....

121 \ _ 1

9

I ~

.... '- ••

6

31 .... la..c ....... "" , ""=

• 2~;~ compressive strain

A. 1';, compressive strain

• Tensile srrength

Plasti c I im it = 32

Liquid limit = 65

I ~J -i?1:- I. u2S 30 35 40 45 50 55 60 65 70

fv~oisturs' Content, Per Cent

Fig_ 11 - Tensile end Unconfined Compression Strength Vs /'/oisture Content (Plastic Index = 33)

0025 1

002,

N 00i5 ..... '-+-

'" ..!J.'

'" on 'lJ ... .....

VI

'-0

~ 00!0 I

00051

0 1

40

1.& 1.2 I _ No Erosion

.. Erosion as inc;icated

.102 204 Erosior rate in

~- I~ 2/h gmj it I ' r

40

"

~~69 --Envelope curve for 35 • no erosion co

"2~ \ •• ::>J

--ftpproximate envei-

\ ope for erosion of I '" 2/h gm/tt / .r

2A

1\ \ 41 \ 3l ~.2. .I i9

II~ - -, \~551 ~!05 1 • • ... A ~ 1 (-, ; ....

~ AJlI,'Mr. -1

16. -• - _I 4.~

--I -5.0 I 45 50 55 60 65 70

I\~oisture Content I Per Cent

Fig. 12- Shear Stress Vs I\/oisture Cont,:mt for Erosion Tests of Duluth Red Clay (Plastic Index = 33)

75

l() [' ...

~

~ ----, (V) M

~ "'" (:J I! ..... ""IIIIIIII

" x

~ (\)

'-0 C

u .... V'> 0

l() 0.... '-0 '---"

~ VI Q)

Q..

E

~ 0

v') ..... c c

~ rJ 0

G U V)

~ '-0 L...

()

<lJ L.

0.... ILl

L.. , 0

~~ ..... <+-c

-<-0 -l-_ C C ,)

0 4-C

l() U 0 l()

Q) U i_

:J ;V + .. L_ V) :J

4-0 If)

~ 0 .'., .:::.

</)

Co > l()

j >.. 4--

V)

C (l)

~~ 0 , M In

~ ",,, .

0)

I.L

L-______________ L-______________ I ________________ L-____________ -J 52 o l() o Lf") c':; -!

o o o , N c.>

i- ~-